1
$\begingroup$

I am running a simple mediation model in a path analysis framework using Mplus.

enter image description here

The direct path in my model is grades (independent variable) predicting happiness (dependent variable). The indirect path is grades predicting happiness through self-esteem. All variables are continuous. There are 3 racial groups in my dataset: White, Black, and Latino. If I wanted to control for race, how would I do that? Would I just run 1 model with all the White participants, and 1 model with all the Black participants? And if I stratify the groups by race (assuming this is how I control for variables), does that mean I have less statistical power in each model?

Improve this question
$\endgroup$
3
  • $\begingroup$ How do you think race should interact in your causal diagram with the other variables? $\endgroup$
    – Adrian Keister
    Commented Feb 2, 2021 at 3:54
  • $\begingroup$ The research corpus suggests a moderate correlation between the independent and dependent variables with race. $\endgroup$
    – Bob Bernstein
    Commented Feb 2, 2021 at 4:50
  • $\begingroup$ What you need to do before answering your question, is to put race as another node in your diagram, and fill in the arrows you think should be there, ESPECIALLY including direction of the arrows. There are possibly other ways to condition for race - if you need to. You won't know if you need to unless you can put it as another node and show its relation to the other nodes. $\endgroup$
    – Adrian Keister
    Commented Feb 2, 2021 at 16:41

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.